CN106709502B - Multi-feature fusion recognition method based on voting method - Google Patents
Multi-feature fusion recognition method based on voting method Download PDFInfo
- Publication number
- CN106709502B CN106709502B CN201611024110.7A CN201611024110A CN106709502B CN 106709502 B CN106709502 B CN 106709502B CN 201611024110 A CN201611024110 A CN 201611024110A CN 106709502 B CN106709502 B CN 106709502B
- Authority
- CN
- China
- Prior art keywords
- voting
- scoring
- feature
- sample
- class
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Character Discrimination (AREA)
Abstract
The invention discloses a multi-feature fusion recognition method based on a voting method, which comprises the following steps of: A. linear expression of the characteristics; B. voting and scoring; C. the invention can solve the problem of lower accuracy in single-sample face recognition by the invention, because the more the voting types are, the more the probability that the voting types are correct simultaneously is, and the higher the accuracy of recognition is.
Description
Technical Field
The invention relates to the technical field of feature recognition, in particular to a multi-feature fusion recognition method based on a voting method.
Background
The single sample face recognition problem is a situation that is often encountered in practice, and therefore, in order for the face recognition technology to be applied to more occasions, the single sample problem must be carefully solved. On the other hand, although only one training sample per person is stored in the face database, which is disadvantageous for most face recognition technologies, several advantages exist that are expected to be achieved in practical applications: (1) The collection of samples is easy, whether indirect or direct: a common component of face recognition systems is the face database, where the construction of a face database storing face images as "templates" is a very time-consuming and labor-consuming task, which is effectively alleviated if only one image is required for each person as training sample. For the same reason, the development of face recognition systems will be easier and, in those cases where direct image acquisition is very difficult, if not impossible, a training sample per person has its unique advantages. Considering monitoring applications in public places, such as airports, stations, etc., a large number of people need to be identified and confirmed. In this case, rather than actually taking an image of each person, we can effectively build a desired face database, such as a passport, identification card, student card, identification card, driver's license, etc., by scanning a photograph on the document. (2) saving storage costs: the storage cost of the face recognition system will be reduced when only one image per person needs to be stored in the database. (3) saving the calculation cost: the computational expense of large-scale applications is significantly reduced because the number of training samples per person directly affects the costs involved in operations such as preprocessing, feature extraction and recognition in the face recognition process.
In summary, in practical applications, the single sample problem is unavoidable and has its unique advantages. In addition, growing acute insight about this particular problem is of greater significance, both in terms of face recognition and in terms of solving other, more general, small sample problems. Therefore, the single-sample face recognition problem has strong academic research value and application value, provides new challenges and new opportunities for the face recognition field, further expands the application range of the face recognition technology by solving the problem, and further improves the related technology.
Disclosure of Invention
The invention aims to provide a multi-feature fusion recognition method based on a voting method, so as to solve the problems in the background technology.
In order to achieve the above purpose, the present invention provides the following technical solutions: a multi-feature fusion identification method based on a voting method comprises the following steps:
A. linear expression of the characteristics;
B. voting and scoring;
C. score classification.
Preferably, the characteristic linear expression in the step a is as follows: firstly, extracting features of a single sample, and if I features are extracted, wherein I is a certain feature extraction method, I is E I; the feature is expressed linearly and the expression is noted as:wherein y is i The ith feature of the picture to be measured; />The ith feature of the training sample picture L; />The expression weight of the L sample under the ith characteristic.
Preferably, the voting scoring mode in the step B is as follows: voting and scoring are carried out on each sample respectively; the scoring method expression is as follows:wherein (1)>Representing the number of votes obtained by voting the jth sample class under the ith feature; ΔGr is a voting function, which is defined as +.>The number of tickets corresponding to the ranking of the ith feature in the order from big to small; l is the highest number of votes, i.e. when +.>In the first row, the number of tickets obtained +.>The second name is L-1 ticket, and the last name is 1 ticket; then the sum of the scores of the j-th class of samples under all the features is obtained: />
Preferably, the classification method in the step C is as follows: after the scoring sum under all the characteristics is obtained, each class is ordered according to the order of magnitude, and the class with the highest score is used as the final classification result, namely
Compared with the prior art, the invention has the beneficial effects that: the invention can solve the problem of lower accuracy in single-sample face recognition because the more the voting types are, the more the probability that the voting types are correct at the same time is possible under the condition of facing complex face change, so that the accuracy of recognition is higher.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention provides a technical scheme that: a multi-feature fusion identification method based on a voting method comprises the following steps:
A. linear expression of the characteristics;
B. voting and scoring;
C. score classification.
In this embodiment, the characteristic linear expression in the step a is as follows: firstly, extracting features of a single sample, and if I features are extracted, wherein I is a certain feature extraction method, I is E I; the feature is expressed linearly and the expression is noted as:wherein y is i The ith feature of the picture to be measured; />The ith feature of the training sample picture L; />The expression weight of the L sample under the ith characteristic.
In this embodiment, the voting scoring mode in the step B is: voting and scoring are carried out on each sample respectively; the scoring method expression is as follows:wherein (1)>Representing the number of votes obtained by voting the jth sample class under the ith feature; ΔGr is a voting function, which is defined as +.>The number of tickets corresponding to the ranking of the ith feature in the order from big to small; l is the highest number of votes, i.e. when +.>In the first row, the number of tickets obtained +.>The second name is L-1 ticket, and the last name is 1 ticket; then the sum of the scores of the j-th class of samples under all the features is obtained: />
In this embodiment, the classification method in step C is as follows: after the scoring sum under all the characteristics is obtained, each class is ordered according to the order of magnitude, and the class with the highest score is used as the final classification result, namely
Experimental example:
the method was tested in three databases, AR, GT and FP respectively. In the three databases, a front face is selected as a training sample for each person, and the other faces are selected as test samples. The feature selection employs LBP, SIFT, PCA and row correlation (RR) four features.
Test data are as follows:
error rate of classification recognition in case of voting using two features:
error rate of classification recognition in case of voting using three features:
error rate of classification recognition in the case of voting using four features:
from the test results, as the number of features increases, so does the accuracy of their identification. Particularly for faces with relatively complex changes, such as the GT database, the recognition accuracy changes more than other databases. This means that the more the voting types are, the more the probability that the voting types are correct at the same time is, so that the accuracy of recognition is higher.
The beneficial effects of the invention are as follows: the invention can solve the problem of lower accuracy in single-sample face recognition because the more the voting types are, the more the probability that the voting types are correct at the same time is possible under the condition of facing complex face change, so that the accuracy of recognition is higher.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (1)
1. A multi-feature fusion recognition method based on a voting method is characterized in that: the method comprises the following steps:
A. linear expression of the characteristics;
B. voting and scoring;
C. scoring and classifying;
the characteristic linear expression mode in the step A is as follows: firstly, extracting features of a single sample, and if I features are extracted, wherein I is a certain feature extraction method, I is E I; the feature is expressed linearly and the expression is noted as:
wherein yi is the ith feature of the picture to be measured;the ith feature of the training sample picture j; />The expression weight of the jth sample under the ith characteristic;
and B, voting scoring mode is as follows: voting and scoring are carried out on each sample respectively; the scoring method expression is as follows:wherein (1)>Representing the number of votes obtained by voting the jth sample class under the ith feature; ΔGr is a voting function, which is defined as +.>The number of tickets corresponding to the ranking of the ith feature in the order from big to small; l is the highest number of votes, i.e. when +.>In the first row, the number of tickets obtained +.>The second name is L-1 ticket, and the last name is 1 ticket; then the sum of the scores of the j-th class of samples under all the features is obtained: />The classification method in the step C is as follows: after the scoring sum under all the characteristics is obtained, each class is ordered according to the order of magnitude, and the class with the highest score is used as the final classification result, namely
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611024110.7A CN106709502B (en) | 2016-11-18 | 2016-11-18 | Multi-feature fusion recognition method based on voting method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611024110.7A CN106709502B (en) | 2016-11-18 | 2016-11-18 | Multi-feature fusion recognition method based on voting method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106709502A CN106709502A (en) | 2017-05-24 |
CN106709502B true CN106709502B (en) | 2023-06-20 |
Family
ID=58941180
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611024110.7A Active CN106709502B (en) | 2016-11-18 | 2016-11-18 | Multi-feature fusion recognition method based on voting method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106709502B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107729835B (en) * | 2017-10-10 | 2020-10-16 | 浙江大学 | Expression recognition method based on fusion of traditional features of face key point region and face global depth features |
CN113436379B (en) * | 2021-08-26 | 2021-11-26 | 深圳市永兴元科技股份有限公司 | Intelligent voting method, device, equipment and storage medium |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102324038B (en) * | 2011-09-06 | 2014-04-16 | 北京林业大学 | Plant species identification method based on digital image |
JP5616310B2 (en) * | 2011-09-27 | 2014-10-29 | 日本電信電話株式会社 | Image matching apparatus and image matching program |
CN103903004B (en) * | 2012-12-28 | 2017-05-24 | 汉王科技股份有限公司 | Method and device for fusing multiple feature weights for face recognition |
CN103810274B (en) * | 2014-02-12 | 2017-03-29 | 北京联合大学 | Multi-characteristic image tag sorting method based on WordNet semantic similarities |
US9852364B2 (en) * | 2014-03-19 | 2017-12-26 | Hulu, LLC | Face track recognition with multi-sample multi-view weighting |
CN104143088B (en) * | 2014-07-25 | 2017-03-22 | 电子科技大学 | Face identification method based on image retrieval and feature weight learning |
CN105827571B (en) * | 2015-01-06 | 2019-09-13 | 华为技术有限公司 | Multi-modal biological characteristic authentication method and equipment based on UAF agreement |
CN106097360A (en) * | 2016-06-17 | 2016-11-09 | 中南大学 | A kind of strip steel surface defect identification method and device |
-
2016
- 2016-11-18 CN CN201611024110.7A patent/CN106709502B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN106709502A (en) | 2017-05-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106874889B (en) | Multiple features fusion SAR target discrimination method based on convolutional neural networks | |
US10509985B2 (en) | Method and apparatus for security inspection | |
US9594984B2 (en) | Business discovery from imagery | |
CN110969087B (en) | Gait recognition method and system | |
CN109829467A (en) | Image labeling method, electronic device and non-transient computer-readable storage medium | |
CN109858476B (en) | Tag expansion method and electronic equipment | |
Singh et al. | A study of moment based features on handwritten digit recognition | |
JP2014232533A (en) | System and method for ocr output verification | |
CN103761531A (en) | Sparse-coding license plate character recognition method based on shape and contour features | |
CN106096517A (en) | A kind of face identification method based on low-rank matrix Yu eigenface | |
CN107798308B (en) | Face recognition method based on short video training method | |
CN104392246A (en) | Inter-class inner-class face change dictionary based single-sample face identification method | |
CN111476310B (en) | Image classification method, device and equipment | |
CN107977439A (en) | A kind of facial image base construction method | |
CN113111880B (en) | Certificate image correction method, device, electronic equipment and storage medium | |
CN104615998A (en) | Vehicle search method based on multiple views | |
CN106709502B (en) | Multi-feature fusion recognition method based on voting method | |
CN107368803A (en) | A kind of face identification method and system based on classification rarefaction representation | |
CN109993049A (en) | A kind of video image structure analysis system towards intelligent security guard field | |
CN110580507B (en) | City texture classification and identification method | |
CN112396060B (en) | Identification card recognition method based on identification card segmentation model and related equipment thereof | |
CN102521623A (en) | Subspace-based incremental learning face recognition method | |
CN115984968A (en) | Student time-space action recognition method and device, terminal equipment and medium | |
CN115375959A (en) | Vehicle image recognition model establishing and recognizing method | |
CN104778479B (en) | A kind of image classification method and system based on sparse coding extraction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |